MM-Snowball Benchmark

What is MM-Snowball?

MM-Snowball is a benchmark for diagnosing hallucination snowballing in multimodal multi-turn dialogue, with fine-grained analysis. It includes data and code…

Released

2026-05-30

Evaluates

Cybersecurity, Multimodal

Openness

unknown

Importer

Claire Radar

Review status

unreviewed

Reported scores

0

Source provenance

• Original evidence https://arxiv.org/abs/2606.00622

MM-Snowball paper

• Paper https://arxiv.org/abs/2606.00622
• Website https://frenkie-chiang.github.io/MM-Snowball

Which sources cite MM-Snowball?

• MM-Snowball

Related Cybersecurity, Multimodal benchmarks

• 3DCodeBench
• A-Bench
• A-OKVQA
• ActionAtlas
• ActiveFly-Bench
• AdversaBench
• Agent Planning Benchmark
• AgentHazard